AI learns yoga pose families, delivering real-time feedback for digital rehab
A novel AI system capable of recognizing yoga poses with high accuracy could pave the way for more effective digital coaching tools, rehabilitation platforms and movement-monitoring applications.
A new artificial intelligence (AI) system developed by researchers from the University of East London can identify yoga poses with over 93% accuracy, according to a recent study published in Scientific Reports. The study, co-authored by Dr. Laura Vanderbloemen, explored the use of four AI models, with Hierarchical CoAtNet 1 emerging as the most effective. This model utilizes a hierarchical learning approach, recognizing broader pose families before diving into specific variations, much like how humans categorize movement.
The researchers found that the AI model processed images swiftly, taking 16-17 milliseconds per image batch and maintaining a real-time streaming speed of 65 to 70 frames per second. Such speeds could facilitate real-time feedback for yoga instructors, physical therapists, and health care professionals, aiding in improving posture quality and movement patterns.
The potential applications of this AI technology extend to digital coaching platforms, rehabilitation tools, and movement-monitoring applications, making health and well-being resources more accessible to individuals who may face challenges in accessing in-person instruction due to location, mobility issues, or cost. By harnessing AI, computer vision, and robotics, this development could significantly expand the reach of health and well-being technologies for a broader range of people.
Written by urgent.news from Medical Xpress's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
